Facilitating machine learning using remote data

Abstract

Techniques and solutions are described for facilitating the use of machine learning techniques. In some cases, a system suitable for providing a machine learning analysis can be different from a remote computer system on which training data for a machine learning model is located. A machine learning task can be defined that includes an identifier for at least one data source on the remote computer system. Data for the at least one data source is received from the remote computer system. At least a portion of the data is processed using a machine learning algorithm to provide a trained model, which can be stored for later use. Data on the remote computing system can be unstructured or structured. Particularly in the case of structured data, a remote computer system can make updated data available to the machine learning task.

What is claimed is: 
     
       1. A computing system comprising:
 memory; 
 one or more hardware processing units coupled to the memory; and 
 one or more computer readable storage media storing instructions that, when executed, cause the computing system to perform operations comprising:
 at the computing system, defining a machine learning task to provide a definition of the machine learning task, the machine learning task using training data located on one or more remote computing systems; 
 at the computing system, including in the definition of the machine learning task an identifier for at least one artefact of a virtual data model on a first remote computing system of the one or more remote computing systems, the at least one artefact of a virtual data model comprising a selection of data from one or more artefacts of the first remote computing system and (1) an annotation indicating that data extraction using the at least one artefact of a virtual data model has been enabled, or (2) an annotation indicating a change propagation technique; 
 at the computing system, sending a data request to the first remote computing system, wherein:
 (1) when the at least one artefact of the virtual data model comprises an annotation indicating that data extraction using the at least one artefact of the virtual data model has been enabled, the first remote computing system processes the data request, determines that the annotation indicates that data extraction has been enabled, and requests data from the one or more artefacts of the first remote computing system at least in part using the selection of data; or 
 (2) when the at least one data artefact of the virtual data model comprises an annotation indicating a change propagation technique, the first remote computing system determines that a change to data identified by the selection of data has occurred in at least one artefact of the one or more artefacts of the first remote computing system and sends changed data to the computing system according to the change propagation technique; 
 
 at the computing system, receiving data for the at least one artefact of the virtual data model from the first remote computing system; 
 at the computing system, processing at least a portion of the received data using a machine learning algorithm to provide a trained model; and 
 at the computing system, storing the trained model. 
 
 
     
     
       2. The computing system of  claim 1 , wherein the at least one artefact of the virtual data model specifies at least one attribute of at least one artefact of the one or more artefacts that should be monitored for changes in a date or timestamp associated with values of the at least one attribute in the at least one artefact. 
     
     
       3. The computing system of  claim 1 , wherein the at least one artefact of the virtual data model is associated with at least one relational database artefact, and a trigger is placed on the at least one relational database artefact in response to the annotation indicating a change propagation technique. 
     
     
       4. The computing system of  claim 1 , wherein receiving data for the at least one artefact of the virtual data model comprises retrieving data from a queue maintained by the first remote computer system. 
     
     
       5. The computing system of  claim 4 , wherein the queue is associated with a table tracking data read by the first computing system. 
     
     
       6. The computing system of  claim 1 , wherein the machine learning task is defined using a machine learning scenario, the machine learning scenario comprising the identifier for the at least one data artefact of the virtual data model and an identifier for a machine learning algorithm to be used to process data from the at least one artefact of the virtual data model. 
     
     
       7. The computing system of  claim 6 , wherein the machine learning scenario further comprises an identifier for an inference processor to be used in analyzing results provided using the trained model. 
     
     
       8. The computing system of  claim 1 , wherein the data is persisted on the computing system. 
     
     
       9. The computing system of  claim 1 , wherein the data is used on the fly in the processing the at least a portion of the received data. 
     
     
       10. The computing system of  claim 9 , wherein the data is not persisted by the computing system after using the data on the fly. 
     
     
       11. The computing system of  claim 1 , wherein the data is first data received at a first time, the operations further comprising:
 receiving second data for the at least one data artefact of the virtual data model from the first remote computer system at a second time, the second data consisting of data changed as compared with the first data. 
 
     
     
       12. The computing system of  claim 1 , the operations further comprising:
 receiving a request to use the trained model, the request comprising input data; 
 processing the input data using the trained model to provide a result; and 
 returning the result in response to the request. 
 
     
     
       13. The computing system of  claim 12 , wherein the request is received from the first remote computing system. 
     
     
       14. A method, implemented in a computing system comprising a memory and one or more processors, comprising:
 at the computing system, defining a machine learning task to provide a definition of the machine learning task, the machine learning task using training data located on one or more remote computing systems; 
 at the computing system, including in the definition of the machine learning task an identifier for at least one artefact of a virtual data model on a first remote computing system of the one or more remote computing systems, the at least one artefact of a virtual data model comprising a selection of data from one or more artefacts of the first remote computing system and (1) an annotation indicating that data extraction using the at least one artefact of the virtual data model has been enabled, or (2) an annotation indicating a change propagation technique;
 at the computing system, sending a data request to the first remote computing system, wherein:
 (1) when the at least one artefact of the virtual data model comprises an annotation indicating that data extraction using the at least one artefact of the virtual data model has been enabled, the first remote computing system processes the data request, determines that the annotation indicates that data extraction has been enabled, and requests data from the one or more artefacts of the first remote computing system at least in part using the selection of data; or 
 (2) when the at least one data artefact comprises an annotation indicating a change propagation technique, the first remote computing system determines that a change to data identified by the selection of data has occurred in the at least one artefact of the one or more artefacts of the first remote computing system and sends changed data to the computing system according to the change propagation technique; 
 
 
 at the computing system, receiving data for the at least one artefact of the virtual data model from the first remote computing system; 
 at the computing system, processing at least a portion of the received data using a machine learning algorithm to provide a trained model; and 
 at the computing system, storing the trained model. 
 
     
     
       15. The method of  claim 14 , wherein the machine learning task is defined using a machine learning scenario, the machine learning scenario comprising the identifier for the at least one data artefact of the virtual data model and an identifier for a machine learning algorithm to be used to process data from the at least one artefact of the virtual data model. 
     
     
       16. The method of  claim 15 , wherein the machine learning scenario further comprises an identifier for an inference processor to be used in analyzing results provided using the trained model. 
     
     
       17. One or more computer-readable storage media storing:
 computer-executable instructions that, when executed by a first computing system comprising at least one hardware processor and at least one memory coupled to the at least one hardware processor, cause the first computing system to define a machine learning task to provide a definition of the machine learning task, the machine learning task using training data located on one or more remote computing systems; 
 computer-executable instructions that, when executed by the first computing system, cause the first computing system to include in the definition of the machine learning task an identifier for at least one artefact of a virtual data model on a first remote computer system of the one or more remote computing systems, the at least one artefact of a virtual data model comprising a selection of data from one or more artefacts of the first remote computing system and (1) an annotation indicating that data extraction using the at least one artefact of the virtual data model has been enabled, or (2) an annotation indicating a change propagation technique;
 computer-executable instructions that, when executed by the first computing system, cause the first computing system to send a data request to the first remote computing system, wherein:
 (1) when the at least one artefact of the virtual data model comprises an annotation indicating that data extraction using the at least artefact of the virtual data model has been enabled, the first remote computing system processes the data request, determines that the annotation indicates that data extraction has been enabled, and requests data from the one or more artefacts of the first remote computing system at least in part using the selection of data; or 
 (2) when the at least one data artefact comprises an annotation indicating a change propagation technique, the first remote computing system determines that a change to data identified by the selection of data has occurred in the at least one artefact of the first remote computing system and sends changed data to the computing system using the change propagation technique; 
 
 
 computer-executable instructions that, when executed by the first computing system, cause the first computing system to receive data for the at least one artefact of the virtual data model from the first remote computing system; 
 computer-executable instructions that, when executed by the first computing system, cause the first computing system to process at least a portion of the received data using a machine learning algorithm to provide a trained model; and 
 computer-executable instructions that, when executed by the first computing system, cause the first computing system to store the trained model.